An assessment of degree-2 Stokes coefficients from Earth rotation data

Variations in the degree-2 Stokes coefficients C20, C21 and S21 can be used to understand long and short-term climate forcing. Here, we derive changes in these coefficients for the period 2003 January–2012 April using Earth rotation data. Earth rotation data contain contributions from motion terms (the effects of winds and currents) and contributions from the effects of mass redistribution. We remove the effects of tides, atmospheric winds and oceanic currents from our data. We compare two different models of atmospheric and oceanic angular momentum for removing the effects of winds and currents: (1) using products from the National Centers for Environmental Prediction and (2) using data from the European Centre for Medium-range Weather Forecasts (ECMWF). We assess the quality of these motion models by comparing the two resulting sets of degree-2 Stokes coefficients to independent degree-2 estimates from satellite laser ranging (SLR), GRACE and a geophysical loading model. We find a good agreement between the coefficients from Earth rotation and the coefficients from other sources. In general, the agreement is better for the coefficients we obtain by removing winds and currents effects using the ECMWF model. In this case, we find higher correlations with the independent models and smaller scatters in differences. This fact holds in particular for C20 and C21, whereas we cannot observe a significant difference for S21. At the annual and semiannual periods, our Earth rotation derived coefficients agree well with the estimates from the other sources, particularly for C21 and S21. The slight discrepancies we obtain for C20 can probably be explained by errors in the atmospheric models and are most likely the result of an over-/underestimation of the annual and semiannual contributions of atmospheric winds to the length-of-day excitation.